[HSC05-P02] Application of sequentially discounting autoregressive (SDAR) on seismic event detection for CO2 injection safety management
The primary task for ATLS management is the micro-seismic event identification in a long-term sustained time series. It is a challenge for ATLS to detect such weak energy seismic signal with potential low Signal to Noise Ratio (SNR). In order to optimize the effect of ATLS seismic monitoring, a high-sensitivity micro-seismic detection method is necessary. In this study, the Sequentially Discounting AutoRegressive (SDAR) method is applied for seismic event detection. Considering that the seismic observation data is a stochastic time series consisting of valid seismic event signal and random noise, the SDAR method assumes that the segments between earthquake and noise are stationary as different autoregressive statistical models. Since the SDAR method avoids the signal energy calculation, it can work on weak seismicity identification, especially micro-seismic events.
Here we apply the SDAR method on the Ocean Bottom Cable (OBC) baseline recording at Tomakomai CCS project for natural earthquake detection. The corresponding data was observed from Feb. 1st 2015 to Jan. 31st 2016. The earthquake detection result confirms that the SDAR is a feasible and robust tool for earthquake detection in long-term time series record, especially for micro-seismic events (magnitude < 1) at CCS site. Near Tomakomai CCS site, only weak earthquakes (magnitude < 2) occurred during the baseline period. This detecting result can be used by ATLS to obtain the background seismic activity information.